def test_select_1(self):
gini = Gini()
s = SelectBestFeatures(method=gini, k=1)
data2 = s(self.titanic)
best = max((gini(self.titanic, f), f)
for f in self.titanic.domain.attributes)[1]
self.assertEqual(data2.domain.attributes[0], best)
def test_continuous_scores_on_discrete_features(self):
data = Impute(self.auro_mpg)
with self.assertRaises(ValueError):
UnivariateLinearRegression(data)
d1 = SelectBestFeatures(method=UnivariateLinearRegression)(data)
self.assertEqual(len(d1.domain), len(data.domain))
def test_select_1(self):
data = Table('titanic')
gini = Gini()
s = SelectBestFeatures(method=gini, k=1)
data2 = s(data)
best = max((gini(data, f), f) for f in data.domain.attributes)[1]
self.assertEqual(data2.domain.attributes[0], best)
def test_select_threshold(self):
data = Table('wine')
anova = ANOVA()
t = 30
data2 = SelectBestFeatures(method=anova, threshold=t)(data)
self.assertTrue(
all(anova(f, data) >= t for f in data2.domain.attributes))
def test_defaults(self):
fs = SelectBestFeatures(k=3)
data2 = fs(Impute(self.auro_mpg))
self.assertTrue(all(a.is_continuous for a in data2.domain.attributes))
data2 = fs(self.wine)
self.assertTrue(all(a.is_continuous for a in data2.domain.attributes))
data2 = fs(self.titanic)
self.assertTrue(all(a.is_discrete for a in data2.domain.attributes))
def test_defaults(self):
fs = SelectBestFeatures(k=3)
data2 = fs(Impute(Table('auto-mpg')))
self.assertTrue(all(isinstance(a, ContinuousVariable) for a in data2.domain.attributes))
data2 = fs(Table('wine'))
self.assertTrue(all(isinstance(a, ContinuousVariable) for a in data2.domain.attributes))
data2 = fs(Table('titanic'))
self.assertTrue(all(isinstance(a, DiscreteVariable) for a in data2.domain.attributes))
def test_defaults(self):
fs = SelectBestFeatures(k=3)
data2 = fs(Impute(Table('auto-mpg')))
self.assertTrue(all(a.is_continuous for a in data2.domain.attributes))
data2 = fs(Table('wine'))
self.assertTrue(all(a.is_continuous for a in data2.domain.attributes))
data2 = fs(Table('titanic'))
self.assertTrue(all(a.is_discrete for a in data2.domain.attributes))
def test_select_threshold(self):
anova = ANOVA()
t = 30
data2 = SelectBestFeatures(method=anova,
threshold=t)(self.heart_disease)
self.assertTrue(
all(
anova(self.heart_disease, f) >= t
for f in data2.domain.attributes))
def test_select_2(self):
gini = Gini()
# 100th percentile = selection of top1 attribute
sel1 = SelectBestFeatures(method=gini, k=1.0)
data2 = sel1(self.titanic)
best = max((gini(self.titanic, f), f)
for f in self.titanic.domain.attributes)[1]
self.assertEqual(data2.domain.attributes[0], best)
# no k and no threshold, select all attributes
sel2 = SelectBestFeatures(method=gini, k=0)
data2 = sel2(self.titanic)
self.assertEqual(len(data2.domain.attributes),
len(self.titanic.domain.attributes))
# 31% = selection of top (out of 3) attributes
sel3 = SelectBestFeatures(method=gini, k=0.31)
data2 = sel3(self.titanic)
self.assertEqual(len(data2.domain.attributes), 1)
# 35% = selection of top (out of 3) attributes
sel3 = SelectBestFeatures(method=gini, k=0.35)
data2 = sel3(self.titanic)
self.assertEqual(len(data2.domain.attributes), 1)
# 1% = select one (out of 3) attributes
sel3 = SelectBestFeatures(method=gini, k=0.01)
data2 = sel3(self.titanic)
self.assertEqual(len(data2.domain.attributes), 1)
# number of selected attrs should be relative to number of current input attrs
sel3 = SelectBestFeatures(method=gini, k=1.0)
data2 = sel3(self.heart_disease)
self.assertEqual(len(data2.domain.attributes), 13)
def test_discrete_scores_on_continuous_features(self):
c = self.iris.columns
for method in (Gini, Chi2):
d1 = SelectBestFeatures(method=method)(self.iris)
expected = \
(c.petal_length, c.petal_width, c.sepal_length, c.sepal_width)
self.assertSequenceEqual(d1.domain.attributes, expected)
scores = method(d1)
self.assertEqual(len(scores), 4)
score = method(d1, c.petal_length)
self.assertIsInstance(score, float)
def test_mixed_features(self):
data = Table('auto-mpg')
data.X = Imputer().fit_transform(data.X)
s = SelectBestFeatures(method=UnivariateLinearRegression(), k=2)
data2 = s(data)
self.assertEqual(
sum(1 for f in data2.domain.attributes
if isinstance(f, ContinuousVariable)), 2)
self.assertEqual(
sum(1 for f in data2.domain.attributes
if isinstance(f, DiscreteVariable)),
sum(1 for f in data.domain.attributes
if isinstance(f, DiscreteVariable)))
def test_discrete_scores_on_continuous_features(self):
c = self.iris.columns
for method in (Gini(), Chi2()):
d1 = SelectBestFeatures(method=method)(self.iris)
expected = \
(c.petal_length, c.petal_width, c.sepal_length, c.sepal_width)
self.assertSequenceEqual(d1.domain.attributes, expected)
scores = method(d1)
self.assertEqual(len(scores), 4)
score = method(d1, c.petal_length)
self.assertEqual(score.ndim, 0) # a scalar
self.assertTrue(np.issubdtype(score.dtype, float))
def test_error_when_using_regression_score_on_classification_data(self):
s = SelectBestFeatures(method=UnivariateLinearRegression(), k=3)
with self.assertRaises(ValueError):
s(self.wine)
def test_error(self):
data = Table('wine')
s = SelectBestFeatures(method=UnivariateLinearRegression(), k=3)
with self.assertRaises(ValueError):
s(data)